| Data Analysis in SPSS [message #954] |
Mon, 02 December 2013 17:17  |
EvisH
Messages: 8
Registered: August 2013
Location: Arlington
|
Member |
|
|
I am working with the 2005 and 2012 datasets for Cote d'Ivoire.
1) How do I get the same numbers as the ones in the published reports for HIV prevalence by region, for example? My numbers differ by a little bit when compared to the ones on the reports.
2) Do I need to use the Complex Samples procedures that you have suggested to other users in order to get the same numbers? I have read the instructions you have indicated on how to set this up but the instructions are not clear. In SPSS, under Analyze, there is no option for complex samples. I am using version 19 of SPSS. I have looked under all the option under Analyze but I do not see how I can set up the Complex Sample procedure.
3)Once I set up this procedure, I run all my syntaxes under this plan?
Thank you very much for a clarifying this issues.
|
|
|
|
| Re: Data Analysis in SPSS [message #966 is a reply to message #954] |
Fri, 06 December 2013 10:34  |
Trevor-DHS
Messages: 805
Registered: January 2013
|
Senior Member |
|
|
First, I moved your post from the "Dataset Use in Stata" forum to the "Dataset Use in SPSS" forum as it is about using data in SPSS.
In response to your questions:
1) To match the numbers in the DHS reports, first check that you are getting the same denominator as DHS. For women and men, the denominator is all de facto women and men who were tested for HIV - check for HIV03 in the range 0:7 for each man or women. The weight to use is HIV05/1000000. Once you are matching the denominator, then the numerator is a subset of these cases, with HIV03 = 1 or 3 (we only report those cases positive with HIV-1).
2) Complex Samples procedures are not required to match the estimates of HIV prevalence. The Complex Samples procedures are necessary if you are producing sampling errors, confidence intervals or testing for significance. Using Complex Sample does not affect the basic calculation of the indicator. If you do not use Complex Samples procedures, your software is assuming that you are using a simple random sample and all of your estimates of sampling errors, confidence intervals and your significance tests would be wrong. Complex Samples is an additional module that needs to be ordered for SPSS. If you have not ordered this module, then it will not appear in your list of available procedures in SPSS.
3) Once you set up a Complex Samples plan for your analysis, you then use this plan with the Complex Samples procedures such as CSDESCRIPTIVES, CSTABULATE, CSLOGISTIC, etc. to run your analysis. These procedures differ a little from the regular SPSS procedures that you may be familiar with, but generally you can produce the equivalent results with the Complex Samples procedures, but taking into account the sample design.
I hope this helps.
|
|
|
|
|
|
The DHS Program User Forum
Members
Search
Help
Register
Login
Home
